Fluently-v4-LCM / README.md
ehristoforu's picture
Update README.md
2e64bc5 verified
metadata
license: other
license_name: fluently-license
license_link: https://huggingface.co/spaces/fluently/License
library_name: diffusers
pipeline_tag: text-to-image
base_model: latent-consistency/lcm-lora-sdv1-5
tags:
  - safetensors
  - stable-diffusion
  - sd1.5
  - LCM
  - fluently
inference:
  parameters:
    num_inference_steps: 8
    guidance_scale: 1.5
    negative_prompt: >-
      (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong
      anatomy, extra limb, missing limb, floating limbs, (mutated hands and
      fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting,
      blurry, amputation

Fluently V4.0-LCM (FAST) - one model for all tasks (Fluently XL)

preview

This model takes images very quickly thanks to LCM technology.

I would like to introduce my model - Fluently! This model was made by merging. I crossed a lot of checkpoints and loras.

About this model

In a nutshell, I took a few checkpoints and a bunch of Loras, I crossed through an extension in AUTOMATIC1111 - SuperMerger (available in extensions). Many factors were taken into account such as: eye quality, correct anatomy, reductions in required promt for a good result.

What makes my model different from others

  • Correct anatomy: my model has the correct anatomy

  • Face details: my model generates beautifully detailed faces and eyes even without AfterDetailer

  • Artistic: my model activates beautiful artistry when it needs it

  • Inpainting: my model is pretty good at Inpainting/Outpainting and I don't need to put in a specially designed model

  • Anime & Comic Book style: my model can draw great Anime and Comic Book art

Merge details

Below is a severely truncated list of models and loras of this model merging:

Models:

  • Juggernaut Final

  • Deliberate V2

  • RPG

  • Realistic Vision V1.3

  • DreamDrop V1

and more models...

Loras:

  • LowRA V2

  • Intricate Details

  • Detail Slider

  • Xeno Detailer

and more loras...

How to use this model

Quick Start

  1. Install this model and start the AUTOMATIC1111

  2. Select this checkpoint

  3. Generate images!

Optimal Parameters

  • Steps: 4-8

  • Sampler: Euler a

  • CFG Scale: 1.2-1.5

  • CLIP-skip: 1

  • Negative Prompt: practically unnecessary

Addion for this model

style.csv for this model: click